The efficacy and safety of tranexamic acid in lumbar surgery: A meta-analysis of randomized-controlled trials.

OBJECTIVES: This meta-analysis aims to assess tranexamic acid (TXA) effectiveness and safety in lumbar surgery. PATIENTS AND METHODS: Renewals of randomized-controlled trials (RCTs) were conducted utilizing databases of medical literature such as PubMed, China Science and Technology Journal Database, Cochrane Library, China National Knowledge Infrastructure (CNKI), and EMBASE to compare principal and safety endpoints. The risk ratio (RR), standard mean difference (SMD), and 95% confidence intervals (CIs) were calculated. For the evaluation of the quality of the included studies, the Cochrane risk of bias criteria were utilized by two authors.
RESULTS: In total, 49 articles were enrolled that included 4,822 patients. Of the patients, 2,653 were administered TXA and 2,169 were in the control group. The findings indicated that TXA was capable of significantly lowering postoperative blood loss (PBL), transfusion rate, transfusion volume, total blood loss (TBL), intraoperative blood loss (IBL), and drainage compared to the control group. Besides, hemoglobin (Hb) and hematocrit (Hct) values were higher in the TXA group compared to the control group. As the safety endpoints, TXA significantly reduced D-dimer levels compared to the control group; however, both TXA and control groups had no significant variations in deep venous thrombosis (DVT). Subgroup analysis was administrated according to the administration method of TXA and the operation type and intravenous and topical TXA were combined in the meta-analysis.
CONCLUSION: This meta-analysis showed that TXA had the potential to significantly lower PBL, transfusion rate, transfusion volume, TBL, IBL, and drainage compared to the control group. Besides, Hb and Hct values were higher in the TXA group compared to the control group. Its hemostatic potential after lumbar spine surgery is trustworthy. It is still controversial in safety endpoints that TXA can significantly reduce D-dimer compared to the control group, without no significant variations in DVT in both the TXA and control groups.

Introduction

With the improvement of human life expectancy and changes in lifestyle, the number of patients undergoing lumbar surgery for lumbar diseases is increasing.[1] In lumbar surgery, extensive muscle tissue stripping can lead to spinal canal decompression, bone graft fusion, a large wound, and internal fixation may injure the long segment of the intraspinal venous plexus, and these factors are expected to cause more postoperative bleeding.[2] The amount of pre- and postoperative bleeding is closely related to the complexity of the operation and is directly related to the time of the drainage tube removal and the need for postoperative blood transfusion.[3] Reducing the exudation of incision blood and removing the drainage tube as soon as possible is not only necessary for postoperative rehabilitation, but it is also essential to minimize the risk of lower-extremity deep venous thrombosis (DVT).[4] Concurrently, minimizing quantity of blood loss from the wound following the surgery may help to eliminate the necessity for a blood transfusion.[5] Therefore, reducing perioperative blood loss is important to ensure the safety of surgery.[6]

To minimize perioperative blood loss, physicians have utilized a variety of techniques, such as controlled hypotension, blood dilution, autologous blood transfusion, and application of hemostatic drugs.[7] Currently, in orthopedic surgery, hemostatic medications with various hemostatic routes have been widely utilized, but due to the need for immobilization, DVT risk exists and, therefore, the application of hemostatic drugs is still controversial.[8] As a common hemostatic drug, tranexamic acid (TXA) is a lysine synthetic derivative and an antifibrinolytic agent.[9] Its pharmacological action is to bind competitively to the lysine binding sites on the source of fibrinolytic enzyme, tissue type plasminogen activator, and plasmin to prevent the dissolution of thrombi.[10-12] Numerous studies have reported that TXA has no effect on enhancing the incidence of DVT, but most of them are routinely used for chemical thromboprophylaxis and, thus, the risk of thrombosis is still not clear.[13-16]

The application of TXA in lumbar surgery is relatively common, but it is still controversial and ambiguous about its safety and effectiveness.[17] Therefore, our study aimed to discover the safety and effectiveness of TXA in lumbar surgery to reinforce the hemostatic medicines clinical application.

Patients and Methods

Search strategy

A literature search utilizing the Preferred Reporting Items for Systematic Reviews and MetaAnalyses (PRISMA) guidelines were carried out by two authors[18] for papers assessing the safety and effectiveness of TXA in lumbar operation. We searched in PubMed, the Cochrane Library, EMBASE, the China National Knowledge Infrastructure (CNKI), and the China Science and Technology Journal Database (commonly known as “VIP”) comprehensively for randomized-controlled trials (RCTs). The language choice is restricted to English or Chinese, and the date of publication was set to begin on January 1st, 2003 to June 30th, 2021. “Tranexamic Acid” and “Lumbar” were utilized as key words, Other meta-analyses and reviews were used to retrieve additional relevant literature. For incomplete or missing data, we contacted the original research authors through electronic mail. Two authors reviewed the retrieved literature. In case of disagreements, a third author was invited to review the paper and render a final decision. No written consent or ethical approval was required, as all data in this meta-analysis were derived from previously published research.

Inclusion and exclusion criteria

Inclusion criteria:

1. The study was an RCT;

2. Evaluated TXA effectiveness and safety in lumbar surgery;

3. The subjects of study were patients who underwent lumbar surgery;

4. On basis of TXA, at least one of groups were assessed;

5. TXA had no dosage or use restrictions;

6. Language options were restricted to English or Chinese;

7. The papers included give sufficient data for analysis.

Exclusion criteria:

1. Animal experiments;

2. Non-randomized trials or semi-randomized controlled trials;

3. Case reports, non-clinical trials, or series;

4. Papers containing wrong or missing data or articles from which data could not be collected.

Endpoints

Total blood loss (TBL) and transfusion rate were the initial endpoints of the study. Secondary endpoints were postoperative drainage, transfusion volume, intraoperative blood loss (IBL), postoperative blood loss (PBL), hemoglobin (Hb) and hematocrit (Hct). Safety endpoints were DVT, and D-dimer (a fibrin degradation product that is traditionally used as a biomarker of DVT).[19,20]

Data extraction

The retrieved studies contents were reviewed by two authors independently. A third author validated the primary endpoints derived by the two authors. The following information were included in extracted data: author's first name, publication year, country conducted in, body mass index (BMI), the size of the sample, sex ratio, intervention, average age, operation type, follow-up time and the endpoints computed in each study. If the study's contents required clarifying, the study's primary author was called up. Conflicts were resolved via prevailing opinion or by calling up a third author who ultimately took the decision.

Risk of bias assessments

Two authors independently appraised the studies' methodological quality using the Cochrane risk of bias criteria. Each item was classified as having a low risk, a high risk, or no obvious risk. The guideline for detecting limitations of this study (risk of bias) in Cochrane Reviews is shown in Table I, along with the corresponding GRADE evaluation of the quality of evidence. Every trial's bias assessment checklist involved seven items: randomization sequence generation, allocation concealment, blinding of participants and personnel, findings appraisal blinding, inadequate data findings, selective reporting, as well as additional biases. We evaluated publication bias according to the guidance shown in Table II.

Statistical analysis

The individual study results were analyzed and pooled using the Stata version 12.0 software (Stata Corp., College Station, TX, USA). Risk ratios (RRs), standardized mean differences (SMDs), and 95% confidence intervals (CIs) with two-sided p values were estimated in the pooled results. A p value of <0.05 was considered statistically significant. The I2 test was utilized to assess heterogeneity. In case of I2 <50%, heterogeneity was deemed to be minor; but, in case of I2 >50%, heterogeneity was deemed to be substantial. If the I2 was <50%, the fixed-effects model was utilized; when the I2 was >50%, the random-effects model was utilized. If more than 10 studies were involved in the analysis of this endpoint, a funnel plot was constituted to scrutinize publication bias, as well as discovering the heterogeneity sources. We conducted a subgroup analysis of the indicators including the patients’ TBL, transfusion rates, DVT, and D-dimer level. Subgroup analyses were performed on the basis of administration and operation type.

Results

Studies retrieved and characteristics

Relying on the (PRISMA) guideline, a total of 2,963 studies were registered. The research titles and abstracts were reviewed to preclude studies that were not pertinent. Then, we excluded research that were not suitable by scanning articles full text. Ultimately, relying on the inclusion and exclusion criteria, 49 studies[21-69] including a total of 4,822 patients were enrolled (Figure 1).[70] At length, 2,653 patients (55.0%) and 2,169 patients (45.0%) were allotted to the experimental and the control groups, respectively. The researches involved were all RCTs in the meta-analysis. The participants’ baseline characteristics in the RCTs are fully revealed in Table III.

Literature quality evaluation

Since studies included were all RCTs, two authors were saddled with the responsibility of assessing the retrieved studies quality relying on the Cochrane risk of bias criteria. In 49 studies, random sequence generation and allocation concealment were performed. Twenty-four studies verified participant and personnel blinding, while 24 studies demonstrated outcome assessment blinding. Other biases were not mentioned in any of the studies. Table IV summarizes the quality score of the literature.

Primary effective endpoints

Total blood loss (mL)

Total blood loss was reported in 27 studies (35 trial comparisons). In all, 2,841 patients were assessed for TBL, with 1,623 and 1,218 allotted to the experimental and the control groups, respectively. The results demonstrated that the control groupʼs TBL were significantly higher than that of the experimental group (SMD: -1.15, 95% CI: -1.37 to -0.92, I2 =87.9%, p=0.000) (Figure 2). We utilized the random-effects model.

Transfusion rate (%)

There were 14 studies (18 trial comparisons) covered the transfusion rate. In all, 172 of 1,366 individuals in the experimental group required blood transfusion, and 337 of 1,039 individuals in the control group required concurrently. The results indicated that TXA significantly reduced blood transfusions incidence compared to the control group (12.6% vs. 31.4%) (RR: 0.41, 95% CI: 0.34 to 0.49 I2 =1.7%, p=0.434) (Figure 3). The fixed-effects model was done.

Secondary effective endpoints

Transfusion volume (mL)

Thirteen studies (14 trial comparisons) reported the transfusion volume. In the aggregate, 1,136 patients were contained to evaluate the transfusion rate, 588 and 548 in the experimental and control groups, respectively. Based on findings, compared to the control group, transfusion volume seems to be significantly lower in the experimental group's transfusion volume (SMD: -2.42, 95% CI: -3.24 to -1.60, I2 =96.9%, p=0.000) as illustrated in Supplementary Figure 1. The random-effects model was done.

Intraoperative blood loss (mL)

Thirty-nine studies (51 trial comparisons) reported IBL. The number of patients was 3,881, with 2,180 allotted to the experimental group and 1,701 to the control group. The statistical findings revealed that, compared to the control group, the experimental group's IBL was significantly lower (SMD: -0.83, 95% CI: -1.05 to -0.61, I2 =91.3%, p=0.000) as illustrated in Supplementary Figure 2. The random-effects model was done.

Postoperative blood loss (mL)

Eleven studies (15 trial comparisons) reported PBL. A total of 1,385 patients were evaluated for PBL. Of them, 761 and 624 were allotted to the experimental and control groups, respectively. Compared to the control group, the experimental PBL was significantly lower (SMD: -2.13, 95% CI: -2.68 to -1.57, I2 =94.9%, p=0.000) as illustrated in Supplementary Figure 3. The random-effects model was done.

Postoperative drainage (mL)

Thirty-five studies (44 trial comparisons) reported postoperative drainage. In all, 3,109 patients were assessed for postoperative drainage, with 1,704 and 1,405 allotted to the experimental and control groups, respectively. The findings revealed that the experimental had significantly lower postoperative drainage than the control group (SMD: -1.55, 95% CI: -1.83 to -1.26, I2 =92.2%, p=0.000) as illustrated in Supplementary Figure 4. The random-effects model was done.

Hemoglobin (g/dL)

Thirty-two studies (42 trial comparisons) reported Hb content. A total of 3,326 patients were involved to evaluate Hb content, of whom 1,863 were in the experimental group and 1,463 in the control group. The findings indicated that the experimental group's Hb content was significantly higher than the control group (SMD: 0.53, 95% CI: 0.36 to 0.71, I2 =83.9%, p=0.000) as illustrated in Supplementary Figure 5. The random-effects model was done.

Hematocrit (%)

Eighteen studies (24 trial comparisons) reported Hct. A total of 1,844 patients were evaluated for Hct, of whom 1,052 and 792 in the experimental and control groups, respectively. The results demonstrated that the TXA group had a greater level of Hct than the control group (SMD: 0.39, 95% CI: 0.08 to 0.70, I2 =91.0%, p=0.000) as illustrated in Supplementary Figure 6. The random-effects model was done.

Safety endpoints

Deep venous thrombosis

Eight studies (11 trial comparisons) covered DVT, of which 19 out of 581 in the experimental group and 23 out of 432 in the control group experienced DVT. There was no significant variation among the TXA and control groups (3.2% vs. 5.3%) (RR: 0.78, 95% CI: 0.48 to 1.28, I2 =0.0%, p=0.926) as illustrated in Figure 4. The fixed-effects model was done.

D-dimer (mg/L)

The concentration of D-dimer can be used in blood tests to help to diagnose thrombosis. Negative results can rule out thrombosis, while positive results suggest thrombosis probability, even so other potential reasons were not excluded. Therefore, its fundamental usage is to exclude thromboembolic diseases with a low probability. D-dimer was evaluated in 19 studies (24 trial comparisons), enrolling 1,837 participants for D-dimer assessment. The experimental group composed of 1,014 participants, whereas the control group composed of 823 participants. The results revealed that, compared to the control group, the experimental group's D-dimer levels were significantly lower (SMD: -0.35, 95% CI: -0.70 to -0.01, I2 =92.5%, p=0.000) as illustrated in Figure 5. The random-effects model was done.

Publication bias and sensitivity analysis and subgroup analysis

According to the TXA administration method and the type of operation, subgroup analysis was done. Subgroup analysis results are listed in Supplementary Figures 7-14. The patients’ TBL in the posterior lumbar surgery (PLS) group, posterior lumbar interbody fusion (PLIF) group, other operative type group and PLIF/transforaminal lumbar interbody fusion (TLIF) group was significantly lower compared to the control group (SMD: -0.84, 95% CI: -1.39 to -0.28, I2 =89.4%, p=0.000; SMD: -1.14, 95% CI: -1.42 to -0.86, I2 =79.9%, p=0.000; SMD: -1.21, 95% CI: -1.67 to -0.75, I2 =92.0%, p=0.000; SMD: -1.94, 95% CI: -2.32 to -1.56, I2 =87.9%, p=0.799) as illustrated in Supplementary Figure 7. The findings revealed that patients’ TBL in the intravenous administration group, topical application group, and intravenous administration before the operation group was significantly lower compared to the control group, whereas there were no significant variations in the intravenous administration group + topical application group (SMD: -1.06, 95% CI: -1.32 to -0.81, I2 =87.7%, p=0.000; SMD: -1.46, 95% CI: -1.95 to -0.97, I2 =87.8%, p=0.000; SMD: -1.55, 95% CI: -2.09 to -1.01; SMD: -0.29, 95% CI: -0.97 to 0.39) as illustrated in Supplementary Figure 8. The transfusion rates in the PLIF group, the other operational type group, as well as the PLIF/TLIF group were all significantly lower compared to the control group. There were no significant variations between the TLIF group and the control group (RR: 0.40, 95% CI: 0.33 to 0.48, I2 =18.3%, p=0.269; RR: 0.60, 95% CI: 0.39 to 0.92, I2 =0.0%, p=0.855; RR: 0.27, 95% CI: 0.11 to 0.70, I2 =0.0%, p=0.658; RR: 0.11, 95% CI: 0.01 to 2.01) as illustrated in Supplementary Figure 9. The transfusion rates in the topical application group, intravenous administration group, and intravenous administration before the operation group were all significantly lower compared to the control group, whereas there were no significant variations while comparing intravenous administration with topical application (RR: 0.40, 95% CI: 0.30 to 0.54, I2 =0.0%, p=0.440; RR: 0.41, 95% CI: 0.32 to 0.53, I2 =0.0%, p=0.684; RR: 0.21, 95% CI: 0.06 to 0.69) as illustrated in Supplementary Figure 10 There were no significant variations in DVT in the PLIF group patients, other operative type group, transforaminal thoracic interbody fusion (TTIF), TLIF, and the control group (RR: 1.00, 95% CI: 0.29 to 3.41, I2 =0.0%, p=0.764; RR: 0.47, 95% CI: 0.13 to 1.64, I2 =0.0%, p=0.805; RR: 0.98, 95% CI: 0.53 to 1.80; RR: 0.33, 95% CI: 0.04 to 3.10) as illustrated in Supplementary Figure 11. There were no significant variations in the intravenous administration group's DVT patients, topical application group, and control group (RR: 0.85, 95% CI: 0.50 to 1.46, I2 =0.0%, p=0.856; RR: 0.54, 95% CI: 0.15 to 1.94, I2 =0.0%, p=0.719) as illustrated in Supplementary Figure 12. Patients in the PLS group had significantly lower D-dimer levels compared to the control group (SMD: -0.31, 95% CI: -0.53 to -0.09, I2 =0.0%, p=0.698) as illustrated in Supplementary Figure 13; however, there were no significant variations in the patients’ D-dimer levels in the PLIF group, other operative type group, TLIF group, and control group (SMD: -0.26, 95% CI: -1.05 to 0.53, I2 =96.0%, p=0.000; SMD: -0.55, 95% CI: -1.11 to 0.02, I2 =92.1%, p=0.000; SMD: 0.11, 95% CI: -0.19 to 0.41, I2 =0.0%, p=0.703). The findings revealed that the patients’ D-dimer levels in the topical application group and intravenous administration before the operation were all significantly lower than those in the control group (SMD: -0.88, 95% CI: -1.61 to -0.15, I2 =94.7%; SMD: -1.45, 95% CI: -1.99 to -0.92, p=0.000), but there were no significant differences among the intravenous administration group, intravenous administration + topical application group and control group (SMD: 0.10, 95% CI: -0.29 to 0.49, I2 =89.9%, p=0.000; SMD: -0.07, 95% CI: -0.49 to 0.36) as illustrated in Supplementary Figure 14.

As demonstrated in Supplementary Figures 15-24, the funnel plot revealed that the retrieved articles had a publication bias. The sensitivity analysis findings are illustrated in Supplementary Figures 25-30.

Discussion

This meta-analysis demonstrated that TXA had a significant impact on lowering the TBL, transfusion rate, transfusion volume, IBL, PBL, drainage and D-dimer compared to the control group and it did not increase the occurrence of DVT; therefore, its effectiveness and safety were proven to be well established.

Recently, with the maturity of lumbar surgical techniques and the improvement of surgical equipment, bleeding during lumbar surgery has been effectively controlled.[71] However, lumbar surgery is still one of the surgical procedures that causes extensive blood loss and, thus, surgeons are concerned about how to reduce perioperative blood loss.[72] The TXA has been approved by the United States Food and Drug Administration (FDA) for more than 30 years and was added to the World Health Organization (WHO) Essential Drugs List in 2011.[73] It shows excellent tolerance, with only rare dose-dependent adverse reactions, including nausea, vomiting, diarrhea, headache, upright reaction, blurred vision, and vertigo.[74] Many original studies and reviews have suggested that TXA is safer than placebo and does not increase the incidence of DVT or pulmonary embolism.[75] Additionally, clinical findings indicate that TXA usage in cardiac valve replacement and total hip arthroplasty can significantly minimize intraoperative blood transfusion volumes without enhancing the risk of thrombosis.[76] Even so, TXA's effectiveness and safety in lumbar surgery still remain controversial.

Currently, there are many articles studying TXA in lumbar surgery. In terms of its efficacy, Du and Feng[77] conducted a meta-analysis to show that TXA had an important ability to minimize IBL and length of hospital stay following lumbar spinal fusion surgery. According to Lu et al.,[78] TXA usage significantly decreased perioperative blood loss and the needs of red blood cell transfusions, but other surgical and clinical outcomes were not significantly different. On the other hand, some scholars put an opposed opinion that TXA might be incapable to reduce blood transfusion rate. Gong et al.[79] performed a meta-analysis and concluded that intravenous TXA had the ability to significantly minimize surgical blood loss. However, TXA treatment did not result in a significant reduction in the transfusion rates in treated patients. Endres et al.[80] performed a retrospective, case-control study and suggested that, when TXA was used in PLS, the Hb concentration was higher and the amount of blood loss was reduced. It lacked the capability to demonstrate a variation in transfusion rates. Furthermore, the safety of the TXA is also under study and some have offered the opinion that it has not any effect on enhancing thrombotic events risk. Bai et al.[81] performed a meta-analysis and proposed that TXA can minimize Hb loss, TBL, intraoperative and PBL, and it does not enhance thrombotic events risk following posterior lumbar fusion. However, there was no significant variation in blood transfusion rates. A retrospective, non-randomized, case-cohort study was performed by Sun et al.[82] and reported that TXA efficiently lowered perioperative blood loss, tube drainage durations, and length of hospitalization and it had no impact on increasing the risk of complications. Ren et al.[83] also carried out a retrospective, casecontrol study and concluded that TXA significantly minimized PBL, shortened the time to withdrawal of drainage tubes and the length of hospitalization in patients receiving PLS fusion surgery, although it did not increase the complication incidence. In contrast, Baldus et al.[84] conducted a comparative study with controls and found that the TXA group had less blood loss and received fewer blood transfusions than the aprotinin treatment group without any significant differences in the intraoperative or postoperative complications. As a result, it is yet unclear if TXA is safe and effective enough to be utilized in the clinic.

Our results revealed that TXA might significantly reduce TBL, transfusion rate, transfusion volume, IBL, PBL, drainage, and D-dimer compared to the control group. While comparing to the control group, TXA could significantly improve Hb and Hct and there were no significant variations in DVT among the TXA group and the control group. We did subgroup and sensitivity analyses after assessing that the endpoints had a high degree of heterogeneity. There were no restrictions on the usages or dose of TXA in our inclusion criteria and, therefore, we performed a subgroup analysis according to the method of administration of TXA (intravenous injection or local injection) and compared their postoperative drainage. Both routes could significantly reduce the patients’ TBL postoperative drainage compared to the control group. Nonetheless, there were no significant variations in postoperative drainage among the two subgroups, and these results cannot explain the heterogeneity. We speculated that this might be because the articles we included had a limited sample size and the patients were relatively heterogeneous. The disunity of the control group and the different dosages used in the TXA group might be also causes of heterogeneity.

To the best of our knowledge, the safety of TXA has been a bigger issue than studies of its efficacy, on account of its hemostatic mechanism that through the abnormal hyperactive fibrinolytic enzyme, causing platelet agglutination and inhibiting the decomposition of coagulation factors, and playing a hemostatic role. Until now, several studies have found that TXA is not associated with the increasing risk of complications; but the patients enrolled in these studies are also routinely prophylaxis with antithrombotic drugs after surgery which may cover the potential increased risk of TXA in venous thromboembolism. Besides, these vast majority of studies also exclude patients with comorbidities and patients who may be at risk for thromboembolism. The result in the meta-analysis suggests that the level of D-dimer decreases in TXA group than the control group. After reviewing the included literatures, we found that, in some of them, the D-dimer levels in the experimental group were somewhat less than in the control group, and there were nonsignificant variations. Others showed that TXA attenuated the increase of D-dimer after surgery. We can speculate that it is related to its anti-fibrinolytic effect: fibrinolytic enzymes, plasminogen, and fibrin binding may be inhibited by TXA by blocking lysine binding sites on plasminogen molecules, thus inhibiting the fibrinolytic decomposition caused by fibrinolytic enzyme. Theoretically, the risk of thrombosis is low after TXA use.

The potential clinical implications are as follows: (i) Thirty RCTs were identified, which comprised 3,042 subjects, more than in previous meta-analyses. The larger, population-inclusive, evidence-based review we conducted summarized the data and might provide a theoretical basis for future clinical drug use; (ii) Subgroup analyses were carried out based on the type of operation and administration route to account for the impact of several parameters on the overall effect; (iii) To determine the source of heterogeneity, we performed a sensitivity analysis to indicate the impact of sample size on the overall effect; and (iv) Ten indicators were assessed including TBL, transfusion rate, transfusion volume, IBL, PBL, drainage Hb, Hct, D-dimer, and DVT, which seemed to be more comprehensive than previous articles. Nonetheless, this study has some limitations: (i) We did not examine the interactions among the subgroup analyses due to the inherent limitations of the enrolled studies; (ii) The impact of the baseline features on the results could not be determined, since the outcome events documented in the enrolled studies were utilized; (iii) As most of the included articles did not report this information, we could not extract relevant data for some baseline features, such as other drug use, hypertension, or diabetes, which may cause some mixed bias. In addition, subgroup analysis according to the dose of TXA, the age of the adults and the safety endpoints, such as the risk of cerebrovascular accident, heart disease, or pulmonary embolism could not be performed; (iv) The outcomes of the various interventions in the control group may show significant heterogeneity. Even so, for ethical issues, we realize that it is unrealistic to compel the original author to refrain from using any hemostatic or anticoagulant interventions; hence, we incorporated all of these articles; (v) Since the limitation of the number of safety events such as cardiac problems or pulmonary embolism in published RCTs, the more safety endpoints could not be included; and (vi) Since there were no obvious findings were found in sensitivity analysis we conducted, it was not detailed in the paper. Moreover, although the results from this meta-analysis did not find an increased risk for DVT, RCTs included almost all exclude patients with comorbidities for this reason and consisted of patients with a low risk. It is still not clear that the safety of TXA in patients with risk factors. Further comprehensive studies with more data are needed to confirm these findings.

In conclusion, this meta-analysis demonstrates that TXA has the potential to significantly minimize TBL, transfusion rate, transfusion volume, IBL, PBL, drainage compared to the control group. Besides, the Hb and Hct values were higher in the TXA group than the control group. Its hemostatic potential after lumbar spine surgery is trustworthy. Besides, it is still controversial in safety endpoints that TXA can significantly reduce D-dimer compared to the control group, whereas there were no significant variations in DVT between the TXA and the control groups.

Table 1

Guidance to assess study limitations (risk of bias) in Cochrane Reviews and corresponding GRADE assessment of quality of evidence

Risk of bias	Across studies	Interpretation	Considerations	GRADE assessment of study limitations
Low	Most information is from studies at low risk of bias.	Plausible bias unlikely to seriously alter the results.	No apparent limitations.	No serious limitations, do not downgrade.
Unclear	Most information is from studies at low or unclear risk of bias.	Plausible bias that about the results.	Potential limitations are unlikely to lower confidence in the estimate of effect.	No serious limitations, do not downgrade.
			Potential limitations are likely to lower confidence in the estimate of effect.	Serious limitations, downgrade one level.
High	The proportion of information from studies at high risk of bias is sufficient to affect the interpretation of results.	Plausible bias that seriously weakens confidence in the results.	Crucial limitation for one criterion, or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect.	Serious limitations, downgrade one level.
			Crucial limitation for one or more criteria sufficient to substantially lower confidence in the estimate of effect.	Very serious limitations, downgrade two levels.

Table 2

Study limitations in randomized-controlled trials: Explanation

	Explanation
Lack of allocation concealment	Those enrolling patients are aware of the group (or period in a crossover trial) to which the next enrolled patient will be allocated (a major problem in “pseudo” or “quasi randomized trials with allocation by day of week, birth date, chart number, etc.).
Lack of blinding	Patient, caregivers, those recording outcomes, those adjudicating outcomes, or data analysts are aware of the arm to which patients are allocated (or the medication currently being received in a crossover trial).
Incomplete accounting of patients and outcome events	Loss to follow-up and failure to adhere to the intention-to-treat principle in superiority trials; or in noninferiority trials, loss to follow-up, and failure to conduct both analyses considering only those who adhered to treatment, and all patients for whom outcome data are available.The significance of particular rates of loss to follow-up, however, varies widely and is dependent on the relation between loss to follow-up and number of events. The higher the proportion lost to follow-up in relation to intervention and control group event rates, and differences between intervention and control groups, the greater the threat of bias.
Selective outcome reporting	Incomplete or absent reporting of some outcomes and not others on the basis of the results.
Other limitations	Stopping trial early for benefit.Substantial overestimates are likely in trials with fewer than 500 events and that large overestimates are likely in trials with fewer than 200 events. Empirical evidence suggests that formal stopping rules do not reduce this bias.

Table 3

Characteristics of studies included in meta-analysis

Authors	Year	Country	Sample size		Female, No (%)		Average age (years)		BMI		Intervention
			E	C	E	C	E	C	E	C	E	C	Follow-up	Operative type	Endpoints
Elmose et al.[21]	2019	Denmark	117	116	58 (0.49)	39 (0.36)	48.9±15.4	51.1±14.9	26.4±3.8	26.2±3.7	TXA,	Equivalent	28 days	Minor lumbar	Total blood loss,
											10 mg/kg, IV	normal saline		spine surgery	Intraoperative blood
											(0.9%), IV				loss, postoperative
															blood loss
Kim et al.[22]	2017	South	24	24	16 (0.67)	9 (0.375)	63.3±7.6	65.2±7.0	NA	NA	TXA, 5 mg/kg,	100 mL normal	7 days	PLIF	Total blood loss,
		Korea									IV, preoperation; saline (0.9%), IV				Intraoperative blood
											A maintenance				loss, postoperative
											dosage of				blood loss,
											1 mg/kg/h, until				postoperative
											5 h after surgery				drainage, Hb, Hct
		Korea									IV, preoperation; saline (0.9%), IV				Intraoperative blood
											A maintenance				loss, postoperative
											dosage of				blood loss,
											2 mg/kg/h, until				postoperative
											5 h after surgery				drainage, Hb, Hct
Liang et al.[23]	2016	China	30	30	15 (0.50)	12 (0.40)	51.1±10.7	53.8±11.2	26.2±4.1	24.9±5.3	Gelfoam was	Gelfoam	3 days	Lumbar spine	Transfusion
											soaked in TXA			surgery	volume,
											(2,000 mg:				postoperative
											20 mL), topical,				drainage, Hb, Hct
											intraopration
Mu et al.[24]	2019	China	45	42	18 (0.40)	19 (0.45)	54.2±7.4	52.6±6.7	24.8±2.0	23.9±1.4	TXA, 15 mg/kg,	Normal saline	84 days	PLIF	Transfusion rate,
											IV, preoperation;	(0.9%),			Intraoperative blood
											A maintenance	100 mL, IV;			loss, postoperative
											dose of 1 mg/kg,	gelatin sponges			blood loss,
											intraoperation	soaked in normal			postoperative
											saline, topical				drainage, Hb, Hct
Mu et al.[24]	2019	China	39	42	17 (0.44)	19 (0.45)	51.8±8.1	52.6±6.7	24.7±1.8	23.9±1.4	Gelfoam was	Normal saline	84 days	PLIF	Transfusion rate,
											soaked in	(0.9%), 100 mL,			Intraoperative blood
											TXA	IV, preoperation;			loss, postoperative
											(1 g: 50 mL),	gelatin sponges			blood loss,
											intraoperation	soaked in normal			postoperative
											saline, topical,				drainage, Hb, Hct
											intraoperation
Nagabhushan et al.[25]	2017	India	25	25	16 (0.64)	14 (0.56)	49.6±9.8	51.7±9.7	NA	NA	TXA, 10 mg/kg,	10 mL normal	NA	Lumbar Spinal	Intraoperative blood
											IV preopration;	saline (0.9%), IV		Fusion Surgery	loss, postoperative
											A maintenance				drainage, Hb
											dosage of
											1 mg/kg/h,
											until closure
Ou et al.[26]	2018	China	59	59	28 (0.47)	29 (0.49)	64.2±4.6	64.0±5.1	22.6±3.3	22.6±3.2	Gelfoam was	Gelfoam was	30 days	Lumbar	Total blood loss,
											soaked in	soaked in		decompression	transfusion volume,
											TXA	normal saline,		and fusion	transfusion rate,
											(1 g: 10 mL),	topical		surgery	postoperative
											topical,				drainage, Hb, Hct,
											intraoperation				D-dimer
Shi et al.[27]	2017	China	50	46	25 (0.50)	24 (0.48)	53.8±12.1	55.8±13.1	NA	NA	TXA, 30mg/kg,	Equivalent	35 days	Posterior lumbar	Total blood loss,
											IV, preoperation;	normal saline		surgery for	transfusion rate,
											A maintenance	(0.9%), IV		stenosis or	Intraoperative blood
											dosage of			spondylolisthesis	loss, postoperative
											2 mg/kg/h, until				drainage, Hb, Hct
											the end of the
											operation
Wang et al.[28]	2013	China	30	30	14 (0.47)	12 (0.4)	63.1±4.0	62.0±4.6	21.7±1.9	22.2±1.9	TXA, 15 mg/kg,	Equivalent	2 days	Posterior	Total blood loss,
											IV, preoperation	normal saline		approach lumbar	intraoperative blood
											(0.9%), IV drip			surgery	loss, postoperative
															blood loss
Wong et al.[29]	2008	Canada	73	74	52 (0.71)	48 (0.64)	56.8±16.2	50.0±16.2	NA	NA	TXA,	Equivalent	90 days	PLIF/PTIF	Transfusion rate,
											10 mg/kg, IV;	normal saline			Hb, D-dimer,
											A maintenance	(0.9%), IV
											dosage
											of 1 mg/kg/h
Xu et al.[30]	2017	China	40	40	21	27	53.1±12	57.4±10.7	25.6±2.8	24.9±3.9	TXA (1g:100ml),	Gelfoam	30 days	Posterior spinal	Total blood loss,
					(0.525)	(0.675)					topical,			fusion surgery	transfusion rate,
											intraoperation				postoperative
															drainage, Hct

Table 4

Assessment of methodological quality of included studies

Study	Random allocation	Hidden distribution	Blind method	Incomplete outcome data	Selective reporting of results	Other bias	Quality grade
Wong et al.[29]	Randomized	No clear	Double-blind	Low	Low	Low	A
Huang and Yang[57]	Randomized	No clear	No clear	Low	Low	Low	C
Wang et al.[28]	Randomized	No clear	Double-blind	Low	Low	Low	B
Bu et al.[51]	Randomized	No clear	Single-blind	Low	Low	Low	B
Huang et al.[56]	Randomized	No clear	No clear	Low	Low	Low	C
Yan[50]	Randomized	No clear	No clear	Low	Low	Low	C
Zhang et al.[55]	Randomized	No clear	No clear	Low	Low	Low	C
Zhang et al.[54]	Randomized	No clear	No clear	Low	Low	Low	C
Liang et al.[23]	Randomized	No clear	No clear	Low	Low	Low	B
Feng[52]	Randomized	No clear	No clear	Low	Low	Low	C
Jia et al.[53]	Randomized	No clear	Double-blind	Low	Low	Low	B
Nian et al.[58]	Randomized	No clear	No clear	Low	Low	Low	C
Shi et al.[27]	Randomized	No clear	Triple-blind	Low	Low	Low	B
Wang et al.[66]	Randomized	No clear	No clear	Low	Low	Low	C
Chang et al.[65]	Randomized	No clear	No clear	Low	Low	Low	C
Kim et al.[22]	Randomized	No clear	Double-blind	Low	Low	Low	B
Nagabhushan et al.[25]	Randomized	No clear	Double-blind	Low	Low	Low	A
Shi[59]	Randomized	No clear	Double-blind	Low	Low	Low	B
Song et al.[64]	Randomized	No clear	No clear	Low	Low	Low	C
Xu et al.[30]	Randomized	No clear	No clear	Low	Low	Low	A
Meng et al.[67]	Randomized	No clear	No clear	Low	Low	Low	C
Zhang and Yang[69]	Randomized	No clear	Double-blind	Low	Low	Low	B
Chen et al.[63]	Randomized	No clear	No clear	Low	Low	Low	C
Hu et al.[61]	Randomized	No clear	No clear	Low	Low	Low	C
Liu and Liu[62]	Randomized	No clear	No clear	Low	Low	Low	C
Mua et al.[24]	Randomized	No clear	No clear	Low	Low	Low	A
Ou et al.[26]	Randomized	No clear	No clear	Low	Low	Low	A
Zhang et al.[60]	Randomized	No clear	Single-blind	Low	Low	Low	B
Elmose et al.[21]	Randomized	No clear	Double-blind	Low	Low	Low	A
Liu et al.[68]	Randomized	No clear	No clear	Low	Low	Low	C
Wang et al.[33]	Randomized	No clear	Double-blind	Low	Low	Low	A
Wang et al.[32]	Randomized	No clear	No clear	Low	Low	Low	C
Deng et al.[31]	Randomized	No clear	Single-blind	Low	Low	Low	B
Xia[37]	Randomized	No clear	Single-blind	Low	Low	Low	B
Xu et al.[34]	Randomized	No clear	Double-blind	Low	Low	Low	A
Yang et al.[35]	Randomized	No clear	Double-blind	Low	Low	Low	A
Zhao et al.[36]	Randomized	No clear	No clear	Low	Low	Low	C
Zhu[38]	Randomized	No clear	Double-blind	Low	Low	Low	A
Ding et al.[47]	Randomized	No clear	Double-blind	Low	Low	Low	A
He et al.[46]	Randomized	No clear	Double-blind	Low	Low	Low	A
Jianjiang et al.[49]	Randomized	No clear	No clear	Low	Low	Low	C
Xia et al.[43]	Randomized	No clear	Double-blind	Low	Low	Low	A
Yang et al.[42]	Randomized	No clear	No clear	Low	Low	Low	C
Yang et al.[48]	Randomized	No clear	No clear	Low	Low	Low	C
Zhang et al.[39]	Randomized	No clear	Single-blind	Low	Low	Low	B
Liu et al.[40]	Randomized	No clear	No clear	Low	Low	Low	C
Mi et al.[44]	Randomized	No clear	No clear	Low	Low	Low	C
Yuan et al.[41]	Randomized	No clear	No clear	Low	Low	Low	C
Zhang et al.[45]	Randomized	No clear	No clear	Low	Low	Low	C